Lean-burn engines have the greatest near term potential for reducing greenhouse gas emissions (25%-35%). However, lean-burn emissions of oxides of nitrogen (NOx), particulate matter (PM), and certain hydrocarbons (HC) species are currently a major concern. Catalytic aftertreatment will be key to the development of lean-burn emissions control technology. Currently, the most promising catalytic materials for lean-burn engines are composed of nanoparticles of precious metals impregnated on g-Al₂ O₃ substrate. The development of new catalytic materials with improved efficiency and durability have been hampered by a gradual loss of efficiency due to coalescence and sintering of these nanoparticle ensembles. Our goal is to use a multi scale simulation approach to determine the mechanisms responsible for nanoparticle coarsening and substrate sintering. This information would be used to determine options for reducing coarsening as a factor in catalyst degradation. An important goal is to define the relationship between the small length and time scale phenomena to macroscopic quantities and use them to establish a clear connection between small scale simulations and continuum simulations . A longer term objective is to relate these phenomena to catalytic performance under realistic conditions.

Teraflop computation is a powerful tool for analyzing, understanding, and predicting the properties of materials because it serves as a bridge between theoretical understanding and experiment. Experiment, theory and computation thus form three synergistic parts of modern scientific research. With computational catalytic science rapidly evolving and growing in importance, it is critical to develop algorithms in conjunction with new theoretical developments, modern computer science approaches, and real experiments.

High operational temperatures and thermal cycling leads to coarsening of supported metal particles as well as sintering of the catalyst support itself. These phenomena contribute greatly to loss of activity in emissions catalysts during service. Coarsening of the metal particles results in the loss of active surface area for the metal. Sintering of the substrate results in reduced surface area for activity. Sintering induced densification and thermal cycling result in stress build up in the substrate that impacts coarsening and can lead to cracking.

Computational Methodology

We propose to develop new simulational capabilities that can be used to provide the most realistic analysis and understanding of these phenomena. A Boundary Integral Method that can accurately and efficiently treat singular structures such as cracks will be used to calculate stress; Level Set, the only viable method for treating discontinuous changes in moving boundaries, will be used for boundary evolution; and Fast Multipole Methods will be used to approximate the far-field boundary integrals, allowing a realistic treatment of the complex topologies. Furthermore, the method can accurately treat both kinetic and thermodynamic processes including sintering, phase transformations, encapsulation of active ingredients, and cracking of the washcoat.

This work will initially be based on extensions of a well-developed approach: first principles techniques based on plane-wave ultra-soft pseudo-potential method, to explore configuration space. This technique provides accurate energetics, forces, and electronic information.

Next, a finite temperature, O(N), k-space, electronic structure code to treat reduced symmetry effects at surfaces that allows systems containing tens of thousands of atoms to be addressed.This new electronic structure method is a multiple scattering based method that makes use of a tight-binding like representation to achieve linear scaling with increasing system size N. This method will allow for both the substrate and surface atoms to be jointly described. The dynamics of a surface under actual in-service use can involve complex competing physiochemical phenomena which can be treated with this new technique.